Sunday, June 12, 2005

Space Crack: Fixing the Turn-based Strategy genre

A discussion of simultaneous turns

Document Note: Each feature in a game design must address a need or solve a problem that occurs with the design.

This part 3 of an ongoing game design document written as a blog. Be sure to catch up on previous posts. In the last installment, we described a typical person who would play SpaceCrack and some of the requirements the must be fullfilled in order for the game to succeed.. This time we'll look at how to reinvent the TBS game genre.

But TBS games suck!
There are good reasons why TBS games are dying as a genre. The fanboy in me hates to face the music, but as a designer I realize that there are fundamental flaws with the reward schedules in a typical TBS game. Here are some common ones:

  • Multiplayer games take forever: In the typical “I go, you go” turn structure you are waiting to play most of the time. Play for 5 minutes and wait for twenty results in lots of bored players.
  • More players results in longer games: As you add players, the wait gets longer. Boring.
  • The first turn is boring: Inevitably nothing happens in the first turn. Then you end your turn and wait for the other player’s results. Often, it is not until you are four or five turns into the game does it get anywhere close to interesting.
  • Complexity rules: As an old genre, there are lots of genre conventions. Most TBS games have loads of statistics and complex game systems. This is comfortable for some advanced players, but acts as a huge entry barrier for newer players.

It is no wonder that people dismiss TBS games. If you are looking for a quick gaming fix, this is the last genre in the world you’d want to play. Luckily, we have a lovely solution to all these issues.

New and improved multiplayer TBS
The genre has been evolving. My friend Lennart Sas created a TBS game called Age of Wonders. It had an original multiplayer turn structure called “simultaneous turns.” SpaceCrack steals his idea because, quite frankly, it is ingenious. Some benefits include:

  • The first few minutes of the game actually exciting.
  • Game play moves faster compared to traditional turn-based games.
  • It works great with both multiplayer and single player.
  • You can have fast games that are similar to an RTS. You can have slow games that are similar to play-by-email. All with the same basic game mechanics.

The turn structure is a variation on real-time networked game play. As far as the player is concerned the universe is always ‘live’ with the state of the playing field maintained on the server.

When the player logs on, they start getting real-time updates on the state of the game world. If another player attacks a planet, the player can watch this happen. No more waiting, no more convoluted phases.

But not real-time…
The game is divided into turns, during which the players can accomplish a fixed number of tasks using a limited number of Command points.

Players spend Command points in order to perform actions ranging from building a new ship to attacking an enemy planet. When the command points are spent, the player’s turn is over.
The turn structure and command points keep the playing field even. You can’t perform twice as many actions as your enemy. You can’t use your twitch skills to pummel an enemy that stepped away from their computer.

Turn Structure Variations

  • New command points every X time period: The player gets new command points every X period of time. For a short game, this could be every 2 minutes. For a long game this could be once a day. If a player does not spend his command points with the turn time limit, the points roll over to the next turn.
  • New command points when all turns are ended: When every player has hit ‘end turn’ or has exhausted their command points, the server gives out new command points. This means games can last forever if there is a laggard.
  • Mixed: If either the time limit or all players have ended their turn, command points are renewed.

Cool technology that needs to be built

  • Server: There is a server that tracks the state of each game. All calculations of world state are performed on the server.
  • Client: There is a client that can display updates from the server when the world state changes. The client must also be able to submit action that change the world state.
  • Command Management System: There is a system that manages command points and the end of turns.


Document Note: Features are often difficult to visualize without a diagram and a game play example. If you want your design to convince, you need to include clear, emotionally exciting examples of each feature in action.

“Ray logs into his latest SpaceCrack game with his regular playing buddies. He got the new turn notification last night in his email, but didn’t get a chance to play since it was date night.
He has 10 command points available. Looks like Porter is online as well and the other guys have already finished their turns. Ray does a quick glance over the map and the news log to see what has occurred.

He moves a few of his ships towards an enemy planet. The missles fly and he destroys the defending ship. Within a minute of playing, he has launched 4 attacks and taken over 1 planet. He spends a couple more points building upgrades on his new planet and another couple of Command points moving some ships in as reinforcements.

A message blinks and Porter sends him a message ‘Hey, check the Troglis system.’ Ray zooms over to the Troglis system just in time to see Porter’s battle ships knock the planet flying. Daniel, the planet’s owner, is asleep in Sweden but that’s okay. The way the command points work, this isn’t like a RTS where he could have mounted some twitch response. Daniel will see the report in the morning.

Ray decides to save his remaining Command points and hits the End Turn button. Once everyone has finished, there’ll be a new round with a batch of fresh Command points waiting him tomorrow. “

I wanted to get the concept of simultaneous turns up front since it has a major impact on the rest of the design. In the next post I'll cover the basic elements of the game design: Token, Verbs, and Rules.

In other news, I've also been mocking up the graphic style of the game and some of the basic interface systems. I'm doing my mockups in Anark Studio and though I'm biased (I designed the damn tool), it is truely a joy to whip up this sort of thing. I banged out a working pie menu in a couple of hours and I didn't need a programmer. As a game designer who can't code my way out of a paper bag, this is truely an empowering experience.

The next post is up! Read it now.

take care


  1. Overall, cool beans. But a question:

    In your example, you say, "Daniel, the planet’s owner, is asleep in Sweden but that’s okay. The way the command points work, this isn’t like a RTS where he could have mounted some twitch response."

    This seems to imply that things happen in some order, and based on the mechanics as I've read them, going first has some advantages. To get back to the example, though Daniel couldn't have performed a "twitch" response, it seems like because his actions this turn/round haven't been spent yet, he will be reacting to the attack on his planet?

    Assuming I'm not way off base here, it seems like you need to add some rules for determining action order.

  2. There will certainly be 'order collision' situations where one person will do something before another person.

    The typical TBS game system says that this situation should be avoided at all cost. A simultaneous turn system admits that it will happen, but then seeks to limit the impact of these situations. Command points help out with this quite a bit, but so does the structure of the environment.

    For example, in an RTS game, a player who leaves his forces alone is screwed. The opponent will come in and destroy the missing player's forces with little defensive resistence.

    In a Simultaneous Turns game, the active player can only do so much damage before his Command Points run out. This is a bit more 'fuzzy' than your traditional TBS game, but in practice it works quite well for a couple of reasons.

    - The board is large: The chances of two players simultaneously having an interesting tactical move that depends on order is quite low. Only rarely do 'order collisions' occur.

    - Going first is not necessarily an advantage: The capture mechanics (which I'll get into in further posts) are such that the attacker is often weaker after the attack than if they stayed a defender. As a result any given tactical situation may or may not have an advantage that goes to the first attacker and neither player can easily predict who will gain the long term upper hand.

    The result is that only rarely (if ever) does a player feel that "If only I had gone first, I would have won the game."

    Admittedly, this is a game design that ends up being 90% turn-based and 10% real-time. That's okay since it still removes the twitch element and allows the drunken monkeys to have a good time.

    When you take the perspective that you aren't trying to make Masters of Magic 2 and instead just want to have a fun game, it can be quite fun to play fast and loose with genre conventions.


  3. Good day.

    I have a suggestion, that might be useful regarding the turn system.
    In the game 'Risk II', a simultaneous turn system is provided as an alternate way to play the game.

    While you issue commands to attack your enemies, you do not know what they are about to do. Your opponents might or might not want to attack the same targets you're about to attack. The orders are given before any attacks are processed.
    What happens after all commands have been issued, is that the attacks are given a certain hierarchy.

    First, two countries attacking each other are processed.
    After that, countries that are attacked by multiple other countries are processed.
    Also, if there are multiple players attacking the same country, if they have defeated the defending country's forces, a 'spoils of war' battle will be held at the end of the turn
    Then, 'normal' attacks are processed.
    Finally, 'spoils of war' battles are processed.

    This allows the game to be fully turn based, while allowing for simultaneous input of orders.

  4. Hi roel,

    From what it sounds like, you are describing a 'we go' style or 'parallel' turn-based system. This was also used in a game called Spaceward Ho!

    Typically such parallel turn based system operate in two phases:
    - Command phase: The users submit a set of command, but none of the commands are processed.
    - Execution phase: Once all the commands are submitted, the commands are executed. If there are conflicts, some form of hierarchical system of rules resolves the conflicts. The results are then returned to the player and everyone can play a new turn.

    I love parallel turn-based systems...they scale very nicely to large numbers of players and you can finish a game within a relatively short amount of time. I originally designed a parallel-turn system but:
    - The core risk/reward cycle takes too long. During the command phase, the player has very limited feedback. They only see results generally after the Execution phase is complete and the results are returned.
    - In short games this can take minutes. In long games, this can take days.
    - For SpaceCrack to be a success, I need feedback to occur within seconds. Make a move and 'bam!' you see the results. Otherwise many players get bored.

    The simultaneous turn system found in Age of Wonders is rather different. It really is pseudo-realtime. When you make a move, you see the results immediately. However, the use of command point really ends up enforcing a turn-like rhythm to the whole experience.

    Great point to bring up. Choosing between linear (I go, you go), parallel (we plan, we go), and simultaneous (We all go until we are spent) was a core design decision that will impact the basic playability of the entire game. At this level, it is well worth looking at the various systems available and weighing the details quite carefully.

    take care

  5. Simultaneous turns was a major feature of CivNet, a multiplayer version of Civilization, that was released by MicroProse back in 1993 and ran on Win3.1.

    The game had such a wide variety of multiplayer modes and switches that you can customize the pace of the game precisely the way you want it.

    I would expect a bit more from someone who claims to know the genre.


  6. Hi Ville

    Thanks for turning me onto the simultaneous turns in CivNet. I have a shocking secret to admit. I primarily played games on the Amiga up until about Windows 95. So a lot of old PC titles I know primarily from friends and my meager explorations.

    Civnet was released to poor reviews/sales so I never got around to digging it back up. Did you enjoy it? I heard that there were some balancing issues in their simultaneous turns mode.

    To be fair to both Sid Meier and the Age of Wonders folks, I'm not overly concerned who originated or polished the idea. I'm more concerned about having it available as a design tool.

    take care

  7. Hi, you might wanna check out Crystallized Time. It's a great turn-based war game.

  8. I'm catching up on your whole (awesome) blog, so I haven't bothered commenting on any of your (really old) articles yet. But if you actually check back to comments a year and a half later, I have something to add/ask: how is this any different from turn-based online RPGs dating back to my BBS days?

    If you've ever played anything from Legend of the Red Dragon all the way up to Travian or Kingdom of Loathing or Domain Knights, you've experience exactly what's described here -- your points/turns/widgets trickle out at a set rate per minute/hour/day. The difference is you're talking about taking it to a smaller scale both in number of players and in time per game. It sounds neat, though.

  9. What you described here is essentially how we got multiplayer to work in Civ4. Have you given that a try?

  10. Hi,

    It sounds like you're underestimating the risk of collisions, since most tactically interesting things happen between two people.

    In a "Command points returned when all players have ended their turn" game I've got a few questions:

    - Isn't there a tactical advantage to going last, since presumably you'd be in a great position to spend next turns command points promptly. Allowing me to hide a vulnerable state created at the end of the first turn. (maybe a cool-down timer if a small subset of users are logged in?)

    - I'm assuming all actions are played in the game universe's present state. So If someone plays all their CP before me, I'll spend my first one at the result of their actions. I will not have the opportunity to interfere with things that have happened. Will I see each intermediate state played back, or only the final state?

    - It feels like being able to have units respond to events automatically would be helpful. Otherwise skirting various defenses could be trivial.

    Interesting writing, I loved TradeWars, and this has me thinking about ways to improve it in the modern age.

  11. See Endless Space for a very good implementation of synchrone turn based strategy game!